Abstract
Spatial vision primarily depends on luminance contrast, but hue variation in spatial patterns (e.g., sinewave gratings) interferes with luminance contrast sensitivity (De Valois & Switkes, 1983). We investigated whether this hue masking effect contributes to people’s interpretations of colors in colormap data visualizations (e.g., weather maps, neuroimaging brain maps). Previous work has shown that interpretations of colormap data visualizations are influenced by the dark-is-more bias: the tendency for people to infer that darker regions map to larger quantities. We investigated whether the magnitude of lightness difference needed to activate the dark-is-more bias depended on hue variation. Participants were presented with colormap data visualizations (8x8 grids) that represented quantities of data measured in different counties. One side of the map was lighter and the other side was darker (left/right balanced over trials). Participants indicated on which side of each colormap (left/right) the measured numbers were larger. The colors in the colormaps either varied only in lightness or varied in both lightness and hue (between-subjects). Overall, participants inferred that darker colors mapped to larger quantities (dark-is-more bias). However, when lightness difference was small (delta L*=5), participants were more likely to make dark-more judgments for colormaps that only varied in lightness, compared to colormaps that varied in lightness and hue. These judgments correlated with darkness difference ratings, in which different participants judged how obvious it was that one of two endpoint colors from the colormap stimuli was darker than the other endpoint (r=.79, p<.05). Our results extend the findings of De Valois & Switkes (1983) to the domain of information visualization—hue variation masked the percept of lightness variation, which in turn weakened activation of the dark-is-more bias. Thus, we must account for hue, not just lightness variation, to understand the effect of the dark-is-more bias on interpretations of colormap data visualizations.